Thioredoxin reductase (TrxR) from Escherichia coli catalyzes the reduction of a disulfide in the protein thioredoxin (Trx) by NADPH. It is presumed that during catalysis TrxR alternates between two conformations, only one of which has been actually observed by x-ray crystallography (Waksman et al., J. Mol. Biol. 236:800, 1994). The second conformation reacts with the substrate Trx. Model building suggests that the two conformations differ by a large rotation (65) of the NADP binding domain relative to the FAD domain. The C135S mutant of TrxR and the C32S mutant of Trx have been crosslinked via a disulfide (Wang etal., Biochemistry 35:4812, 1996). Evidence of Wang et al. suggests that this complex is locked in the second (proposed) conformation, so analysis of this or a related complex would allow the determination of the structure of that conformation. Two crosslinked complexes, C138-C35 and C138-C32,were prepared for crystallization trials. Only the TrxR(C138)-Trx(C32) complex has crystallized reproducibly in a form with adequate resolution. This form could occasionally be indexed on our home instrument (R-AXIS IV) in a primitive orthorhombic cell (a=88, b=158, c=458 ) but data collection were precluded by the inability of the R-AXIS instrument to resolve the long cell edge. The ability of the CHESS F-1 beamline to resolve large unit cell edges was critical for studies of this crystal form. Nine partial datasets were collected at CHESS. The data were measured to resolutions of 2.7 to 3.3 with Rsym values of 3.9-9.4%, but because of rapid crystal decay (even at ultralow temperatures), none of the individual datasets was complete. Two datasets could be merged to produce one almost complete set of intensities indexed in space group P212121. The overall completeness to 3.3 is 95% with an Rmerge of 8.2%. Structures of all the constituents of the crosslinked complex are known, but the analysis is challenging because of the size of the asymmetric unit. Initial cross rotation and tranlation searches used the dimer of FAD domains as a search model and have located these central domains in four of the 7-8 protein dimers which crystal density measurements predict are present in the asymmetric unit. Searches with the NADP domains and with other combinations of fragments, using a variety of molecular replacement alogorithms, are in progress.